← BCM 213 Business Statistics
Simple Linear Regression and Correlation Analysis
Notes
Overview of Bivariate Analysis
While univariate statistics inspect single operational variables in isolation, corporate decision making frequently depends on understanding relationships between paired variables. Business managers assess whether promotional expenditure drives customer turnover, whether employee training hours reduce customer service complaints, or whether fuel prices affect haulage costs. Bivariate analysis supplies the statistical mechanisms to measure association and build predictive models.
Pearsons Correlation Coefficient
Correlation analysis evaluates the degree and direction of linear association between two continuous variables without implying causality. Pearson correlation coefficient, denoted as r, ranges strictly between negative one and positive one. A value of positive one indicates perfect direct linear correlation, zero indicates no linear relationship, and negative one signifies an inverse linear relationship. In commercial interpretation, coefficients between 0.7 and 1.0 generally denote strong association, though analysts must guard against spurious correlation driven by lurking variables.
Simple Linear Regression Model
When association is confirmed, regression analysis models the mathematical dependence of an outcome variable, denoted Y, on an explanatory variable, denoted X. The simple linear regression equation is written as Y equals a plus b times X. In this model, a represents the intercept, which is the baseline value of Y when X is zero, while b represents the regression slope, which indicates the expected change in Y per unit increase in X. Ordinary Least Squares technique estimates these parameters by minimizing the sum of squared vertical residuals between actual observations and the fitted regression line.
Assessing Model Fit with R-Squared
The coefficient of determination, denoted R-squared, is obtained by squaring the Pearson correlation coefficient. It measures the proportion of total variation in the dependent variable that is explained by the independent variable in the regression line, providing an immediate index of predictive reliability.
Practical example
A retail apparel boutique in Westlands, Nairobi, tracks monthly social media advertising budget in thousands of Kenyan Shillings against total monthly sales in hundreds of thousands of Kenyan Shillings over four representative quarters.
Quarter 1: Advertising 2, Sales 7.
Quarter 2: Advertising 4, Sales 11.
Quarter 3: Advertising 6, Sales 13.
Quarter 4: Advertising 8, Sales 17.
Step 1: Compute summary terms.
Count n equals 4.
Sum of X equals 2 plus 4 plus 6 plus 8, which is 20. Mean of X equals 5.
Sum of Y equals 7 plus 11 plus 13 plus 17, which is 48. Mean of Y equals 12.
Sum of product X times Y equals 14 plus 44 plus 78 plus 136, totaling 272.
Sum of squared X equals 4 plus 16 plus 36 plus 64, totaling 120.
Step 2: Calculate the regression slope b.
Numerator equals n times sum of XY minus sum of X times sum of Y.
Numerator equals 4 times 272 minus 20 times 48, which is 1088 minus 960, yielding 128.
Denominator equals n times sum of squared X minus square of sum of X.
Denominator equals 4 times 120 minus 20 squared, which is 480 minus 400, yielding 80.
Slope b equals 128 divided by 80, which is 1.6.
Step 3: Calculate the intercept a.
Intercept a equals mean of Y minus b times mean of X.
Intercept a equals 12 minus 1.6 times 5, which is 12 minus 8, equaling 4.
Step 4: Formulate model and forecast.
The regression equation is Sales equals 4 plus 1.6 times Advertising.
If the Westlands boutique invests 10 thousand Kenyan Shillings in advertising next month, projected sales will be 4 plus 1.6 times 10, resulting in 20 hundred thousand Kenyan Shillings, or 2000000 Kenyan Shillings.
Quiz
1. What does a Pearson correlation coefficient of negative 0.85 imply between production cost and profit margin?
2. In the estimated regression equation for the Westlands boutique, what does the slope value of 1.6 represent?
3. If the coefficient of determination between training hours and sales revenue is 0.64, how is this interpreted?
Sign in to submit and securely grade your quiz.